Quick start | Examples | Paper | Citation | Contact
A Taichi-based numerical package for high-performance simulations of multiscale and multiphysics geophysical problems. Developed by Multiscale Geomechanics Lab, Zhejiang University.
GeoTaichi is a collection of several numerical tools, currently including Discrete Element Method (DEM), Material Point Method (MPM), Material Point-Discrete element method (MPDEM), and Finite Element Method (FEM), that cover the analysis of the Soil-Gravel-Structure-Interaction in geotechnical engineering. The main components of GeoTaichi is illustrated as follows:
GeoTaichi is a research project that is currently under development. Our vision is to share with the geotechnical community a free, open-source (under the GPL-3.0 License) software that facilitates the relevant computational research. In the Taichi ecosystem, we hope to emphasize the potential of Taichi for scientific computing. Furthermore, GeoTaichi is high parallelized, multi-platform (supporting for Windows, Linux and Macs) and multi-architecture (supporting for both CPU and GPU).
Have a cool example? Submit a PR!
| Column collapse | Dam break | Strip footing |
|---|---|---|
 |
 |
 |
| Granular packing | Debris Flow | Triaxial shear test |
|---|---|---|
 |
 |
 |
The performance of GeoTaichi is compared with similar simulator in the field, such as MUSEN and CoSim, showing a notable improvement in both computational efficiency and memory usage.
| A sphere impacting granular bed | Granular column impacting cubic particles |
|---|---|
 |
 |
- Python packages
| Package name | Version | Features |
|---|---|---|
| python | >=3.8 | Programming language |
| psutil | lastest | Get GPU memory |
| taichi | ==1.6.0 | High performance computing |
| numpy | ==1.23.5 | Pre-processing |
| scipy | ==1.10.1 | Pre-processing |
| pybind11 | ==2.10.4 | Warp C++ code |
| trimesh | ==3.20.1 | Import stl/obj files |
| shapely | ==1.8.0 | Delaunay triangulation |
- Change the current working directory to the desired location and download the GeoTaichi code:
cd /path/to/desired/location/
git clone https://github.com/Yihao-Shi/GeoTaichi
cd GeoTaichi
- Install essential dependencies
# Install python and pip
sudo apt-get install python3.8
sudo apt-get install python3-pip
# Install python packages (recommand to add package version)
bash requirements.sh
- Install CUDA, detailed information can be referred to official installation guide
- Set up environment variables
sudo gedit ~/.bashrc
$ export PYTHONPATH="$PYTHONPATH:/path/to/desired/location/GeoTaichi"
source ~/.bashrc
- Install Anaconda
- start Anaconda Prompt
- Navigate to a folder where geotaichi_env.yml is located.
- clone geotaichi as:
git clone https://github.com/Yihao-Shi/GeoTaichi
- run command:
conda env create -f geotaichi_env.yml
- run command:
conda activate geotaichi
- correct the environment (the last part should be modified to the path of geotaichi):
conda env config vars set PYTHONPATH=%PYTHONPATH%;.\path\to\GeoTaichi
- run command:
conda activate geotaichi
- run a benchmark (column collapse):
python DPmaterial
Remark: line 3 of the examples should be modified based on the availability of the GPU. If CPU is available, the following should be used;
init('cpu')
pip install geotaichi
To visualize the VTS files produced by some of the scripts, it is recommended to use ParaView. To visualize the output in ParaView, use the following procedure:
- Open the .vts or .vtu file in ParaView
- Click on the "Apply" button on the left side of the screen
- Make sure under "Representation" that "Surface" or "Surface with Edges" is selected
- Under "Coloring" select variables and the approriate measure (i.e. "Magnitude", X-direction displacement, etc.)
Currently, only the tutorial of DEM in Chinese version is available in doc. Users can set up simulations by specifying numerical parameters and configuring the desired simulation settings in a Python script. More detailed about Python scripts can be found in the example floder.
Discrete element method is a powerful tool to simulate the movement of granular materials through a series of calculations that trace individual particles constituting the granular material.
- Sphere and multisphere particles
- Generating particle packings by specifying initial void ratio or particle number in a box/cylinder/sphere/triangular prism
- Two neighbor search algorithms, brust search/linked-cell
- Two velocity updating schemes, symlectic Euler/velocity Verlet
- Three contact models, including linear elastic, hertz-mindlin and linear rolling
- Supporting plane (infinite plane)/facet (servo wall)/triangle patch (suitable for complex boundary condition)
The material point method (MPM) is a numerical technique used to simulate the behavior of solids, liquids, gases, and any other continuum material. Unlike other mesh-based methods like the finite element method, MPM does not encounter the drawbacks of mesh-based methods (high deformation tangling, advection errors etc.) which makes it a promising and powerful tool in computational mechanics.
- Nine Constitutive Models, including linear elastic/neo-hookean/Von-Mises/isotropic hardening plastic/Mohr-Coulomb/Drucker-Prager/(cohesive) modified cam-clay/Newtonian fluid/Bingham fluid
- Two improved velocity projection techniques, including MLS/APIC
- Three stress update schemes, including USF/USL/MUSL/APIC
- Three stabilization techniques, including mix integration/B-bar method/F-bar method
- Two smoothing mehod, including strain/stress smoothing
- Supporting Dirichlet (Fix/Reflect/Friction)/Neumann boundary conditions
- Supporting total/updating Lagrangian explicit MPM and implicit MPM (Newmark-$\beta$ method)
- Free surface detection
- Two contact models, including linear elastic, hertz-mindlin
- Support DEM-MPM-Mesh contact, feasible simulating complex boundary conditions
- Multilevel neighbor search
- Two way or one way coupling
- Restart from a specific time step
- A simple GUI powered by Taichi
- VTU(Paraview) and NPZ(binary files) files are generated in the process of simualtion
- Supporting force chain visualization
- Developing a well-structured IGA modules
This project is licensed under the GNU General Public License v3 - see the LICENSE for details.
Please kindly star ⭐ this project if it helps you. We take great efforts to develope and maintain it 😁😁.
If you publish work that makes use of GeoTaichi, we would appreciate if you would cite the following reference:
@article{shi2024geotai,
author = {Y. H. Shi, N. Guo., Z. X. Yang},
title = {GeoTaichi: a Taichi-powered high-performance numerical simulator for multiscale geophysical problems},
year = {2024},
journal = {Computer Physics Communications},
pages = {109219},
publisher = {Elsevier},
doi = {https://doi.org/10.1016/j.cpc.2024.109219}
}We thank all amazing contributors for their great work and open source spirit. We welcome all kinds of contributions to file an issue at GitHub Issues.
- If you spot any issue or need any help, please mail directly to [email protected].
V0.2.2 (July 22, 2024)
- Fix computing the intersection area between circles and triangles
- Add "Destory" and "Reflect" boundaries in DEM modules, see examples
V0.2 (July 1, 2024)
- Fix some bugs in DEM and MPM modules, see details
- Add some advanced constitutive model
V0.1 (January 21, 2024)
- First release GeoTaichi
